XSEDE
XSEDE (eXtreme Science and Engineering Discovery Environment) is an NSF-funded virtual organization that integrates and coordinates the sharing of advanced digital services - including supercomputers and high-end visualization and data analysis resources - with researchers nationally to support science.[1]
David Chin is the XSEDE Campus Champion for Drexel. Please contact mailto:urcf-support@drexel.edu for questions regarding XSEDE.
Available Resources
XSEDE offers a wide variety of resources, including CPU compute, GPU compute, and high performance storage.[2]
Startup Allocations
Startup allocations are small allocations meant for benchmarking applications as a prelude to a full XSEDE proposal/application. These are granted on a short timescale, typically within about 10 days of application.
Startup allocations are a good way to see if your particular workflow will perform well on XSEDE resources, and to get an idea of the performance improvement.
Information about XSEDE startup allocations can be found here: https://portal.xsede.org/allocations/startup
Research Allocations
Research allocations are meant for "production" use. Researchers should apply for a startup allocation before submitting a proposal for a Research Allocation.
Information about XSEDE research allocations can be found here: https://portal.xsede.org/allocations/research
Extended Collaborative Support Services (ECSS)
The XSEDE Extended Collaborative Support Service (ECSS) is available to assist your team with problems that require support above and beyond that provided through the XSEDE ticket system. Visit the following URL for more details on ECSS.
Science Gateways
XSEDE also offers Science Gateways[3], which are community-provided interfaces to XSEDE resources. These gateways are typically web-based user interfaces, which provide a convenient way of accessing and utilizing XSEDE resources.
Useful Information
- Video: Writing and Submitting a Successful XSEDE Proposal (Recorded: September 12, 2018; Run time: 1:13:13)
Miscellaneous Tips
XSEDE resources use Slurm[4] as the resource manager and job scheduler. (The equivalent of Univa Grid Engine on Proteus.)
- To submit jobs:
sbatch - To view job status:
squeue - To view job usage statitistics after the job completes:
job_infoJOB_ID(replace JOB_ID by the numerical job ID of the job you are interested in)
References
[2] XSEDE Resources